Performance Architect - AI Hardware

TEEMA; United States; Remote; USD 140,000 - 210,000; Full time

Job summary

A tech company is seeking a Performance Architect to develop frameworks for AI workload modeling and optimization. This remote-friendly role requires expertise in neural network operations and computer architecture. Ideal candidates will have skills in Python and C++ and will work closely with hardware and software teams to enhance performance.

Qualifications

• Deep understanding of neural network operations and AI workload behavior.
• Strong foundation in computer architecture, SoCs, and memory hierarchies.
• Experience building performance models or simulation frameworks.
• Proficiency in Python, C++, and/or spreadsheet-based modeling.
• Familiarity with compiler mapping and hardware constraints.
• Ability to think systematically—from high-level AI models to low-level execution details.

Responsibilities

• Build system-level performance models for AI/ML workloads.
• Analyze model execution flow and map NN operations to hardware.
• Develop tools and automation to evaluate system performance.
• Translate architecture into performance estimates and optimization strategies.
• Collaborate with HW architects and software teams.
• Identify and address bottlenecks in data movement.
• Provide feedback loops to architecture and compiler design based on measured or modeled results.

Skills

Neural network operations

AI workload behavior

Computer architecture

SoCs

Memory hierarchies

Performance modeling

Python

C++

Tools

Architecture simulators

Co-design tools

Job description

Get AI-powered advice on this job and more exclusive features.

Performance Architect – AI Workload Modeling & System Optimization

Location: U.S. or Canada (Remote-Friendly)

Employment Type: Full-Time

Help Build What Could Be the Fastest AI Chip on the Planet

Join a stealth-mode team of semiconductor veterans, architecture experts, and systems engineers creating a custom AI processor that’s already demonstrating performance beyond what's currently on the market. Think: better than Qualcomm, faster than NVIDIA—and you're helping shape the heart of it.

This is not just another accelerator. It’s a radical rethinking of how AI workloads are run at the silicon level—and we’re looking for an AI Performance Architect to bridge the gap between neural networks and the hardware designed to run them.

The Role

You’ll develop the frameworks and tools to simulate, analyze, and optimize how real-world AI models perform on our custom architecture. This role requires a deep understanding of how AI workloads behave—from operations to memory to execution flow—and how to map that behavior into performance models that guide architecture and compiler development.

You’ll be the architect of performance insight, helping us understand how every NN op interacts with our chip and how we can make it faster, leaner, and smarter.

What You’ll Do

• Build system-level performance models for AI/ML workloads using tools like C++, Python, and spreadsheets
• Analyze model execution flow and map neural network operations to custom hardware components
• Develop tools and automation to evaluate system performance across real workloads
• Translate architecture into performance estimates and optimization strategies
• Collaborate closely with HW architects, compiler engineers, and software teams to fine-tune the HW/SW stack
• Identify and address bottlenecks by simulating data movement, memory usage, and compute behavior
• Provide feedback loops to architecture and compiler design based on measured or modeled results

What You Bring

• Deep understanding of neural network operations and AI workload behavior
• Strong foundation in computer architecture, SoCs, and memory hierarchies
• Experience building performance models or simulation frameworks
• Proficiency in Python, C++, and/or spreadsheet-based modeling
• Familiarity with compiler mapping, HW constraints, and architecture exploration
• Ability to think systematically—from high-level AI models to low-level execution details

Bonus Points For:

• Prior work mapping AI models into custom hardware or accelerators
• Experience working with architecture simulators or co-design tools
• Hands-on involvement in early-stage hardware-software optimization projects

Why This Team?

Because they’re not following trends—they’re setting them.

Because the product already crushes benchmarks before hitting the market.

Because the team is smart, humble, and hell-bent on changing what AI compute can be.

And because being part of something this big, this early, only happens once or twice in a career.

Ready to Map the Future of AI Compute?

Apply now to shape the performance layer of one of the most ambitious AI hardware platforms being built today.